72 PROCEEDINGS OF THE AMERICAN ACADEMY 



cent.; and when by 10 cm' added drop by drop, with 10 cm. 8 more 

 subsequent to precipitation (experiments 100 and 101) -|-0.0015 

 g r . 5 or _)_5.58 per cent. The mean error of the determination by 

 precipitating with 20 cm. 3 of non-ammoniacal magnesic chloride mix- 

 ture, in slightly ammoniacal solutions (experiments 102 to 107 and 

 110 to 115) was about -j-0.0018 gr. upon 0.0269 gr. of phosphoric 

 oxide, or -j-6.69 per cent. ; about -f 0.0065 gr. upon 0.2017 gr. of 

 phosphoric oxide, or -J-3.22 per cent., in solutions containing no 

 citrate, and in solutions containing a citrate -(-0.0170 gr., or -(-8.35 

 per cent. The mean error when 20 cm. 8 of nou-ammoniacal magnesic 

 chloride mixture was added to the solution of phosphate and tung- 

 state containing no free ammonia (experiments 108 and 109) 

 amounted to about -(-0.0063 gr. upon 0.0269 gr. of phosphoric oxide, 

 or -(-23.60 per cent. 



The use of citric acid in many of the experiments above, in accord- 

 ance with a suggestion of Dr. Gibbs, appears to be attended with un- 

 favorable results. Thus in experiments (85) to (88), in which the 

 first precipitates were dissolved and reprecipitated, this appears to 

 have lowered the results below those of (89) to (91) ; while in ex- 

 periments (80) to (82) and in experiments (112) to (115), in which 

 the first precipitate was not dissolved, it appears to have raised the 

 results above those of (83) and (84), and (110) and (111) respectively, 

 — facts which are perhaps to be explained by attributing to the citrate 

 both a solvent and contaminating action upon the precipitate. 



Of the experiments upon the phosphomolybdates (123) and (124), 

 in which the precipitation was effected by 1.5 cm. 3 of magnesic 

 chloride mixture, 1.5 cm. 3 of the same being added afterward, 

 show a mean error of less than -j-0.0002 gr. upon 0.026!) gr. of 

 phosphoric oxide, or -4-0.55 per cent. ; and experiments (116) to (118) 

 in which 20 cm. 3 of the same precipitant were employed -[-0.0018 gr. 

 upon 0.0270 gr. of phosphoric oxide, or -(-6.66 per cent; experiments 

 (119) to (1A2), in two of which the precipitation was effected by 

 magnesic sulphate mixture and in two by magnesic chloride mixture, 

 the first precipitate being washed with the precipitant, dissolved in 

 hydrochloric acid and again thrown down, less than -j-0.0002 gr. upon 

 0.0270 gr. of phosphoric oxide, or -J-0.65 per cent. In experiments 

 (125) to (129) in which 20 cm. 3 of the magnesic chloride mixture 

 deprived of ammonia were added to slightly ammoniacal solutions, 

 the mean error was less than -(-0.0003 gr. on 0.0269 gr. of 

 phosphoric oxide, or -(-0.92 per cent., and less than -(-0.0013 gr. on 

 2017 gr. of phosphoric oxide, or -(-0.02 per cent. 



